From efb960df45efb634e2639ff911475a9d598cacee Mon Sep 17 00:00:00 2001
From: eead-csic-compbio <compbio@eead.csic.es>
Date: Thu, 27 Oct 2016 17:27:40 +0200
Subject: [PATCH] fixed bug in plot_matrix_heatmap.sh -c X ; now only rows with
 values < X are retained, as expected

---
 CHANGES.txt            |  1 +
 lib/phyTools.pm        |  8 ++++----
 plot_matrix_heatmap.sh | 10 ++++++----
 3 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/CHANGES.txt b/CHANGES.txt
index 324d253..fa00cec 100644
--- a/CHANGES.txt
+++ b/CHANGES.txt
@@ -155,3 +155,4 @@
 17102016: updated manual with FAQ about ANI calculation on soft-core clusters (thanks Raffael Inglin)
 18102016: added FAQ explaining the meaning of branch numbers in compare_clusters.pl -T trees
 18102016: added extra check to pfam_enrich.pl
+27102016: fixed bug in plot_matrix_heatmap.sh -c X ; now only rows with values < X are retained, as expected (thanks Valerie!)
diff --git a/lib/phyTools.pm b/lib/phyTools.pm
index b87529c..d6d7935 100755
--- a/lib/phyTools.pm
+++ b/lib/phyTools.pm
@@ -281,7 +281,7 @@ sub read_FASTA_sequence
     if(!open(FASTA,"gzip -dc $infile |"))
     {
       die "# read_FASTA_sequence: cannot read GZIP compressed $infile $!\n"
-        ."# please check gzip in installed\n";
+        ."# please check gzip is installed\n";
     }
   }
   elsif($infile =~ /\.bz2$/ || $magic eq "BZ") # BZIP2 compressed input
@@ -289,7 +289,7 @@ sub read_FASTA_sequence
     if(!open(FASTA,"bzip2 -dc $infile |"))
     {
       die "# read_FASTA_sequence: cannot read BZIP2 compressed $infile $!\n"
-        ."# please check bzip2 in installed\n";
+        ."# please check bzip2 is installed\n";
     }
   }
   else{ open(FASTA,"<$infile") || die "# read_FASTA_sequence: cannot read $infile $!\n"; }
@@ -428,7 +428,7 @@ sub read_FASTA_file_array
     if(!open(FASTA,"gzip -dc $infile |"))
     {
       die "# read_FASTA_sequence_array: cannot read GZIP compressed $infile $!\n"
-        ."# please check gzip in installed\n";
+        ."# please check gzip is installed\n";
     }
   }
   elsif($infile =~ /\.bz2$/ || $magic eq "BZ") # BZIP2 compressed input
@@ -436,7 +436,7 @@ sub read_FASTA_file_array
     if(!open(FASTA,"bzip2 -dc $infile |"))
     {
       die "# read_FASTA_sequence_array: cannot read BZIP2 compressed $infile $!\n"
-        ."# please check bzip2 in installed\n";
+        ."# please check bzip2 is installed\n";
     }
   }
   else{ open(FASTA,"<$infile") || die "# read_FASTA_sequence_array: cannot read $infile $!\n"; }
diff --git a/plot_matrix_heatmap.sh b/plot_matrix_heatmap.sh
index 238625a..190497a 100755
--- a/plot_matrix_heatmap.sh
+++ b/plot_matrix_heatmap.sh
@@ -253,13 +253,13 @@ if [ $sim_cutoff_int -ne 100 ]
 then
    heatmap_outfile="${tab_file%.*}_sim_cutoff_${sim_cutoff}_heatmap.$outformat"
    echo "# Plotting file $heatmap_outfile"
+   nj_tree="${tab_file%.*}_sim_cutoff_${sim_cutoff}_BioNJ.ph"
 else
    heatmap_outfile="${tab_file%.*}_heatmap.$outformat"
    echo "# Plotting file $heatmap_outfile"
+   nj_tree="${tab_file%.*}_BioNJ.ph"
 fi
 
-nj_tree="${tab_file%.*}_BioNJ.ph"
-
 # 2) call R using a heredoc and write the resulting script to file 
 R --no-save -q <<RCMD > ${progname%.*}_script_run_at_${start_time}.R
 library("gplots")
@@ -272,8 +272,10 @@ rownames(mat_dat) <- rnames
 
 if($sim_cutoff < 100)
 {
-   rows <- (apply( mat_dat , 1 , function(x) any( x < $sim_cutoff) ) )
-   mat_dat <-  mat_dat[rows, rows]
+   tmp_mat = mat_dat
+   diag(tmp_mat) = NA
+   rows <- (!apply( tmp_mat , 1 , function(x) any( x > $sim_cutoff , na.rm=T) ) )
+   mat_dat <- mat_dat[rows, rows]
 }
 
 if($reorder_clusters > 0){